IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group I: The Physical Science Basis

7.4.2 Nitrogen Compounds

The N cycle is integral to functioning of the Earth system and to climate (Vitousek et al., 1997; Holland et al., 2005a). Over the last century, human activities have dramatically increased emissions and removal of reactive N to the global atmosphere by as much as three to five fold. Perturbations of the N cycle affect the atmosphere climate system through production of three key N-containing trace gases: N2O, ammonia (NH3) and NOx (nitric oxide (NO) + nitrogen dioxide (NO2)). Nitrous oxide is the fourth largest single contributor to positive radiative forcing, and serves as the only long-lived atmospheric tracer of human perturbations of the global N cycle (Holland et al., 2005a). Nitrogen oxides have short atmospheric lifetimes of hours to days (Prather et al., 2001). The dominant impact of NOx emissions on the climate is through the formation of tropospheric ozone, the third largest single contributor to positive radiative forcing (Sections 2.3.6, 7.4.4). Emissions of NOx generate indirect negative radiative forcing by shortening the atmospheric lifetime of CH4 (Prather 2002). Ammonia contributes to the formation of sulphate and nitrate aerosols, thereby contributing to aerosol cooling and the aerosol indirect effect (Section 7.5), and to increased nutrient supply for the carbon cycle (Section 7.5). Ammonium and NOx are removed from the atmosphere by deposition, thus affecting the carbon cycle through increased nutrient supply (Section

Atmospheric concentrations of N2O have risen 16%, from about 270 ppb during the pre-industrial era to 319 ppb in 2005 (Figure 7.16a). The average annual growth rate for 1999 to 2000 was 0.85 to 1.1 ppb yr–1, or about 0.3% per year (WMO, 2003). The main change in the global N2O budget since the TAR is quantification of the substantial human-driven emission of N2O (Table 7.7; Naqvi et al., 2000; Nevison et al., 2004; Kroeze et al., 2005; Hirsch et al., 2006). The annual source of N2O from the Earth’s surface has increased by about 40 to 50% over pre-industrial levels as a result of human activity (Hirsch et al., 2006). Human activity has increased N supply to coastal and open oceans, resulting in decreased O2 availability and N2O emissions (Naqvi et al., 2000; Nevison et al., 2004).

Figure 7.16

Figure 7.16. (a) Changes in the emissions of fuel combustion NOx and atmospheric N2O mixing ratios since 1750. Mixing ratios of N2O provide the atmospheric measurement constraint on global changes in the N cycle. (b) Changes in the indices of the global agricultural N cycle since 1850: the production of manure, fertilizer and estimates of crop N fixation. For data sources see http://www-eosdis.ornl.gov/ (Holland et al., 2005b) and http://www.cmdl.noaa.gov/. Figure adapted from Holland et al. (2005c).

Since the TAR, both top-down and bottom-up estimates of N2O have been refined. Agriculture remains the single biggest anthropogenic N2O source (Bouwman et al., 2002; Smith and Conen, 2004; Del Grosso et al., 2005). Land use change continues to affect N2O and NO emissions (Neill et al., 2005): logging is estimated to increase N2O and NO emissions by 30 to 350% depending on conditions (Keller et al., 2005). Both studies underscore the importance of N supply, temperature and moisture as regulators of trace gas emissions. The inclusion of several minor sources (human excreta, landfills and atmospheric deposition) has increased the total bottom-up budget to 20.6 TgN yr–1 (Bouwman et al., 2002). Sources of N2O now estimated since the TAR include coastal N2O fluxes of 0.2 TgN yr–1 (±70%; Nevison et al., 2004) and river and estuarine N2O fluxes of 1.5 TgN yr–1 (Kroeze et al., 2005). Box model calculations show the additional river and estuarine sources to be consistent with the observed rise in atmospheric N2O (Kroeze et al., 2005).

Top-down estimates of surface sources use observed concentrations to constrain total sources and their spatial distributions. A simple calculation, using the present-day N2O burden divided by its atmospheric lifetime, yields a global stratospheric loss of about 12.5 ± 2.5 TgN yr–1. Combined with the atmospheric increase, this loss yields a surface source of 16 TgN yr–1. An inverse modelling study of the surface flux of N2O yields a global source of 17.2 to 17.4 TgN yr–1 with an estimated uncertainty of 1.4 (1 standard deviation; Hirsch et al., 2006). The largest sources of N2O are from land at tropical latitudes, the majority located north of the equator. The Hirsch et al. inversion results further suggest that N2O source estimates from agriculture and fertilizer may have increased markedly over the last three decades when compared with an earlier inverse model estimate (Prinn et al., 1990). Bottom-up estimates, which sum individual source estimates, are more evenly distributed with latitude and lack temporal variability. However, there is clear consistency between top-down and bottom-up global source estimates, which are 17.3 (15.8–18.4) and 17.7 (8.5–27.7) TgN yr–1, respectively.

Concentrations of NOx and reduced nitrogen (NHx = NH3 + ammonium ion (NH4+)) are difficult to measure because the atmospheric lifetimes of hours to days instead of years generate pronounced spatial and temporal variations in their distributions. Atmospheric concentrations of NOx and NHx vary more regionally and temporally than concentrations of N2O. Total global NOx emissions have increased from an estimated pre-industrial value of 12 TgN yr–1 (Holland et al., 1999; Galloway et al., 2004) to between 42 and 47 TgN yr–1 in 2000 (Table 7.7). Lamarque et al. (2005a) forecast them to be 105 to 131 TgN yr–1 by 2100. The range of surface NOx emissions (excluding lightning and aircraft) used in the current generation of global models is 33 to 45 TgN yr–1 with small ranges for individual sources. The agreement reflects the use of similar inventories and parametrizations. Current estimates of NOx emissions from fossil fuel combustion are smaller than in the TAR.

Since the TAR, estimates of tropospheric NO2 columns from space by the Global Ozone Monitoring Experiment (GOME, launched in 1995) and the SCanning Imaging Absorption SpectroMeter for Atmospheric CHartographY (SCIAMACHY, launched in 2002) (Richter and Burrows, 2002; Heue et al., 2005) provide constraints on estimates of NOx emissions (Leue et al., 2001). Martin et al. (2003a) use GOME data to estimate a global surface source of NOx of 38 TgN yr–1 for 1996 to 1997 with an uncertainty factor of 1.6. Jaeglé et al. (2005) partition the surface NOx source inferred from GOME into 25.6 TgN yr–1 from fuels, 5.9 TgN yr–1 from biomass burning and 8.9 TgN yr–1 from soils. Interactions between soil emissions and scavenging by plant canopies have a significant impact on soil NOx emissions to the free troposphere: the impact may be greatest in subtropical and tropical regions where emissions from fuel combustion are rising (Ganzeveld et al., 2002). Boersma et al. (2005) find that GOME data constrain the global lightning NOx source for 1997 to the range 1.1 to 6.4 TgN yr–1. Comparison of the tropospheric NO2 column of three state-of-the-art retrievals from GOME for the year 2000 with model results from 17 global atmospheric chemistry models highlights significant differences among the various models and among the three GOME retrievals (Figure 7.17, van Noije et al., 2006). The discrepancies among the retrievals (10 to 50% in the annual mean over polluted regions) indicate that the previously estimated retrieval uncertainties have a large systematic component. Top-down estimates of NOx emissions from satellite retrievals of tropospheric NO2 are strongly dependent on the choice of model and retrieval.

Figure 7.17

Figure 7.17. Tropospheric column NO2 from (a) satellite measurements and (b) atmospheric chemistry models. The maps represent ensemble average annual mean tropospheric NO2 column density maps for the year 2000. The satellite retrieval ensemble comprises three state-of-the-art retrievals from GOME; the model ensemble includes 17 global atmospheric chemistry models. These maps were obtained after smoothing the data to a common horizontal resolution of 5° × 5° (adapted from van Noije et al., 2006).

Knowledge of the spatial distribution of NOx emissions has evolved significantly since the TAR. An Asian increase in emissions has been compensated by a European decrease over the past decade (Naja et al., 2003). Richter et al. (2005; see also Irie et al., 2005) use trends for 1996 to 2004 observed by GOME and SCIAMACHY to deduce a 50% increase in NOx emissions over industrial areas of China. Observations of NO2 in shipping lanes from GOME (Beirle et al., 2004) and SCIAMACHY (Richter et al., 2004) give values at the low end of emission inventories. Data from GOME and SCIAMACHY further reveal large pulses of soil NOx emissions associated with rain (Jaeglé et al., 2004) and fertilizer application (Bertram et al., 2005).

All indices show an increase since pre-industrial times in the intensity of agricultural nitrogen cycling, the primary source of NH3 emissions (Figure 7.16b and Table 7.7; Bouwman et al., 2002). Total global NH3 emissions have increased from an estimated pre-industrial value of 11 TgN yr–1 to 54 TgN yr–1 for 2000 (Holland et al., 1999; Galloway et al., 2004), and are projected to increase to 116 TgN yr–1 by 2050.

Table 7.7. Global sources (TgN yr–1) of NOx, NH3 and N2O for the 1990s.


Source NOx NH3 N2O 
Anthropogenic sources            
Fossil fuel combustion & industrial processes 33 (20–24) 25.6 (21–28) 0.3 (0.1–0.5) 2.5d  1.3/0.7 (0.2–1.8) 0.7 (0.2–1.8)d 
Aircraft 0.7 (0.2–0.9) e (0.5–0.8) 
Agriculture 2.3f (0–4) 1.6g 34.2 (16–48) 35g (16–48)  6.3/2.9 (0.9–17.9) 2.8 (1.7–4.8)g 
Biomass and biofuel burning 7.1 (2–12) 5.9 (6–12) 5.7 (3–8) 5.4d (3–8) 0.5 (0.2–1.0) 0.7 (0.2–1.0)g 
Human excreta – – 2.6 (1.3–3.9) 2.6g (1.3–3.9)  – 0.2g (0.1–0.3)h 
Rivers, estuaries, coastal zones – – – – – 1.7 (0.5–2.9)i 
Atmospheric deposition – 0.3g – – – 0.6j (0.3–0.9)h 
Anthropogenic total 43.1 33.4 42.8 45.5 8.1/4.1 6.7 
Natural sources            
Soils under natural vegetation 3.3f (3–8) 7.3j (5–8) 2.4 (1–10) 2.4g (1–10) 6.0/6.6 (3.3–9.9) 6.6 (3.3–9.0)g 
Oceans – – 8.2 (3–16) 8.2g (3–16)  3.0/3.6 (1.0–5.7) 3.8 (1.8–5.8)k 
Lightning 5 (2–12) 1.1–6.4 (3–7) – – – – 
Atmospheric chemistry <0.5 – – – 0.6 (0.3–1.2) 0.6 (0.3–1.2)c 
Natural total 8.8 8.4–13.7 10.6 10.6 9.6/10.8 11.0 
Total sources 51.9 (27.2–60.9) 41.8–47.1 (37.4–57.7) 53.4 (40–70)  56.1 (26.8–78.4) 17.7/14.9 (5.9–37.5) 17.7 (8.5–27.7) 


a Values from the TAR: NOx from Table 4.8 with ranges from Tables 4.8 and 5.2; NH3 from Table 5.2, unless noted.

b Parentheses show the range of emissions used in the model runs described in Table 7.9. See text for explanation. Where possible, the best estimate NOx emission is based on satellite observations. None of the model studies includes the NOx source from oxidation of NH3, which could contribute up to 3 TgN yr–1. The source of NOx from stratosphere-troposphere exchange is less than 1 TgN yr–1 in all models, which is well constrained from observations of N2O-NOx correlations in the lower stratosphere (Olsen et al., 2001).

c Values are from the TAR, Table 4.4; Mosier et al. (1998); Kroeze et al. (1999)/Olivier et al. (1998): a single value indicates agreement between the sources and methodologies of the different studies.

d Van Aardenne et al. (2001), range from the TAR.

e The aircraft source is included in the total for industrial processes. The parentheses indicate values used in model runs.

f The total soil NOx emissions estimate of 5.6 provided in Table 4.8 of the TAR was distributed between agriculture and soil NOx according to the proportions provided in the TAR, Table 5.2.

g Bouwman et al. (2001, Table 1); Bouwman et al. (2002) for the 1990s; range from the TAR or calculated as ±50%.

h Estimated as ±50%.

i Kroeze et al. (2005); Nevison et al. (2004); estimated uncertainty is ±70% from Nevison et al. (2004).

j All soils, minus the fertilized agricultural soils indicated above.

k Nevison et al. (2003, 2004), combining the uncertainties in ocean production and oceanic exchange.

The primary sink for NHx and NOx and their reaction products is wet and dry deposition. Estimates of the removal rates of both NHx and NOx are provided by measurements of wet deposition over the USA and Western Europe to quantify acid rain inputs (Hauglustaine et al., 2004; Holland et al., 2005a; Lamarque et al., 2005a). Chemical transport models represent the wet and dry deposition of NOx and NHx and their reaction products. A study of 29 simulations with 6 different tropospheric chemistry models, focusing on present-day and 2100 conditions for NOx and its reaction products, projects an average increase in N deposition over land by a factor of 2.5 by 2100 (Lamarque et al., 2005b), mostly due to increases in NOx emissions. Nitrogen deposition rates over Asia are projected to increase by a factor of 1.4 to 2 by 2030. Climate contributions to the changes in oxidized N deposition are limited by the models’ ability to represent changes in precipitation patterns. An intercomparison of 26 global atmospheric chemistry models demonstrates that current scenarios and projections are not sufficient to stabilise or reduce N deposition or ozone pollution before 2030 (Dentener et al., 2006).